如题，我这是我的 ~/.codex/config.toml 中的配置，但是在实际使用是，deepseek-v4-pro还是上下文的窗口大小还是256k，是我的配置有问题吗，求各位大佬帮忙看下 🥲 
model_provider = "cpa"
model = "codex/deepseek-v4-pro"
model_reasoning_effort = "xhigh"
disable_response_storage = true
preferred_auth_method = "apikey"
model_context_window = 1000000
model_auto_compact_token_limit = 900000
web_search = "live"